Contributors: 37
Author Tokens Token Proportion Commits Commit Proportion
Yazen Ghannam 2035 38.94% 49 31.01%
Jacob Shin 1145 21.91% 3 1.90%
Borislav Petkov 547 10.47% 25 15.82%
Robert Richter 302 5.78% 4 2.53%
Aravind Gopalakrishnan 298 5.70% 9 5.70%
Thomas Gleixner 180 3.44% 12 7.59%
Muralidhara M K 134 2.56% 3 1.90%
Mike Travis 96 1.84% 2 1.27%
Shirish S 80 1.53% 1 0.63%
Greg Kroah-Hartman 55 1.05% 7 4.43%
Andi Kleen 54 1.03% 5 3.16%
Avadhut Naik 50 0.96% 1 0.63%
Sebastian Andrzej Siewior 35 0.67% 4 2.53%
Ammar Faizi 30 0.57% 1 0.63%
Ingo Molnar 27 0.52% 3 1.90%
Smita Koralahalli 26 0.50% 3 1.90%
Hidetoshi Seto 23 0.44% 2 1.27%
Boris Ostrovsky 18 0.34% 1 0.63%
Chen Yucong 16 0.31% 3 1.90%
Tony Luck 15 0.29% 2 1.27%
Jan Beulich 14 0.27% 2 1.27%
Andrew Morton 11 0.21% 1 0.63%
Julia Lawall 8 0.15% 1 0.63%
Mukul Joshi 5 0.10% 1 0.63%
Qiuxu Zhuo 4 0.08% 1 0.63%
Seiji Aguchi 3 0.06% 1 0.63%
Linus Torvalds (pre-git) 2 0.04% 1 0.63%
Daniel Walter 2 0.04% 1 0.63%
Tejun Heo 2 0.04% 1 0.63%
Rafael J. Wysocki 2 0.04% 1 0.63%
Kees Cook 1 0.02% 1 0.63%
Linus Torvalds 1 0.02% 1 0.63%
Konstantin Khlebnikov 1 0.02% 1 0.63%
Luiz Fernando N. Capitulino 1 0.02% 1 0.63%
Thomas Weißschuh 1 0.02% 1 0.63%
Emese Revfy 1 0.02% 1 0.63%
Dave Hansen 1 0.02% 1 0.63%
Total 5226 158 


// SPDX-License-Identifier: GPL-2.0-only
/*
 *  (c) 2005-2016 Advanced Micro Devices, Inc.
 *
 *  Written by Jacob Shin - AMD, Inc.
 *  Maintained by: Borislav Petkov <bp@alien8.de>
 */
#include <linux/interrupt.h>
#include <linux/notifier.h>
#include <linux/kobject.h>
#include <linux/percpu.h>
#include <linux/errno.h>
#include <linux/sched.h>
#include <linux/sysfs.h>
#include <linux/slab.h>
#include <linux/init.h>
#include <linux/cpu.h>
#include <linux/smp.h>
#include <linux/string.h>

#include <asm/traps.h>
#include <asm/apic.h>
#include <asm/mce.h>
#include <asm/msr.h>
#include <asm/trace/irq_vectors.h>

#include "internal.h"

#define NR_BLOCKS         5
#define THRESHOLD_MAX     0xFFF
#define INT_TYPE_APIC     0x00020000
#define MASK_VALID_HI     0x80000000
#define MASK_CNTP_HI      0x40000000
#define MASK_LOCKED_HI    0x20000000
#define MASK_LVTOFF_HI    0x00F00000
#define MASK_COUNT_EN_HI  0x00080000
#define MASK_INT_TYPE_HI  0x00060000
#define MASK_OVERFLOW_HI  0x00010000
#define MASK_ERR_COUNT_HI 0x00000FFF
#define MASK_BLKPTR_LO    0xFF000000
#define MCG_XBLK_ADDR     0xC0000400

/* Deferred error settings */
#define MSR_CU_DEF_ERR		0xC0000410
#define MASK_DEF_LVTOFF		0x000000F0

/* Scalable MCA: */

/* Threshold LVT offset is at MSR0xC0000410[15:12] */
#define SMCA_THR_LVT_OFF	0xF000

static bool thresholding_irq_en;

struct mce_amd_cpu_data {
	mce_banks_t     thr_intr_banks;
	mce_banks_t     dfr_intr_banks;

	u32		thr_intr_en: 1,
			dfr_intr_en: 1,
			__resv: 30;
};

static DEFINE_PER_CPU_READ_MOSTLY(struct mce_amd_cpu_data, mce_amd_data);

static const char * const th_names[] = {
	"load_store",
	"insn_fetch",
	"combined_unit",
	"decode_unit",
	"northbridge",
	"execution_unit",
};

static const char * const smca_umc_block_names[] = {
	"dram_ecc",
	"misc_umc"
};

#define HWID_MCATYPE(hwid, mcatype) (((hwid) << 16) | (mcatype))

struct smca_hwid {
	unsigned int bank_type;	/* Use with smca_bank_types for easy indexing. */
	u32 hwid_mcatype;	/* (hwid,mcatype) tuple */
};

struct smca_bank {
	const struct smca_hwid *hwid;
	u32 id;			/* Value of MCA_IPID[InstanceId]. */
	u8 sysfs_id;		/* Value used for sysfs name. */
	u64 paddrv	:1,	/* Physical Address Valid bit in MCA_CONFIG */
	    __reserved	:63;
};

static DEFINE_PER_CPU_READ_MOSTLY(struct smca_bank[MAX_NR_BANKS], smca_banks);
static DEFINE_PER_CPU_READ_MOSTLY(u8[N_SMCA_BANK_TYPES], smca_bank_counts);

static const char * const smca_names[] = {
	[SMCA_LS ... SMCA_LS_V2]	= "load_store",
	[SMCA_IF]			= "insn_fetch",
	[SMCA_L2_CACHE]			= "l2_cache",
	[SMCA_DE]			= "decode_unit",
	[SMCA_RESERVED]			= "reserved",
	[SMCA_EX]			= "execution_unit",
	[SMCA_FP]			= "floating_point",
	[SMCA_L3_CACHE]			= "l3_cache",
	[SMCA_CS ... SMCA_CS_V2]	= "coherent_slave",
	[SMCA_PIE]			= "pie",

	/* UMC v2 is separate because both of them can exist in a single system. */
	[SMCA_UMC]			= "umc",
	[SMCA_UMC_V2]			= "umc_v2",
	[SMCA_MA_LLC]			= "ma_llc",
	[SMCA_PB]			= "param_block",
	[SMCA_PSP ... SMCA_PSP_V2]	= "psp",
	[SMCA_SMU ... SMCA_SMU_V2]	= "smu",
	[SMCA_MP5]			= "mp5",
	[SMCA_MPDMA]			= "mpdma",
	[SMCA_NBIO]			= "nbio",
	[SMCA_PCIE ... SMCA_PCIE_V2]	= "pcie",
	[SMCA_XGMI_PCS]			= "xgmi_pcs",
	[SMCA_NBIF]			= "nbif",
	[SMCA_SHUB]			= "shub",
	[SMCA_SATA]			= "sata",
	[SMCA_USB]			= "usb",
	[SMCA_USR_DP]			= "usr_dp",
	[SMCA_USR_CP]			= "usr_cp",
	[SMCA_GMI_PCS]			= "gmi_pcs",
	[SMCA_XGMI_PHY]			= "xgmi_phy",
	[SMCA_WAFL_PHY]			= "wafl_phy",
	[SMCA_GMI_PHY]			= "gmi_phy",
};

static const char *smca_get_name(enum smca_bank_types t)
{
	if (t >= N_SMCA_BANK_TYPES)
		return NULL;

	return smca_names[t];
}

enum smca_bank_types smca_get_bank_type(unsigned int cpu, unsigned int bank)
{
	struct smca_bank *b;

	if (bank >= MAX_NR_BANKS)
		return N_SMCA_BANK_TYPES;

	b = &per_cpu(smca_banks, cpu)[bank];
	if (!b->hwid)
		return N_SMCA_BANK_TYPES;

	return b->hwid->bank_type;
}
EXPORT_SYMBOL_GPL(smca_get_bank_type);

static const struct smca_hwid smca_hwid_mcatypes[] = {
	/* { bank_type, hwid_mcatype } */

	/* Reserved type */
	{ SMCA_RESERVED, HWID_MCATYPE(0x00, 0x0)	},

	/* ZN Core (HWID=0xB0) MCA types */
	{ SMCA_LS,	 HWID_MCATYPE(0xB0, 0x0)	},
	{ SMCA_LS_V2,	 HWID_MCATYPE(0xB0, 0x10)	},
	{ SMCA_IF,	 HWID_MCATYPE(0xB0, 0x1)	},
	{ SMCA_L2_CACHE, HWID_MCATYPE(0xB0, 0x2)	},
	{ SMCA_DE,	 HWID_MCATYPE(0xB0, 0x3)	},
	/* HWID 0xB0 MCATYPE 0x4 is Reserved */
	{ SMCA_EX,	 HWID_MCATYPE(0xB0, 0x5)	},
	{ SMCA_FP,	 HWID_MCATYPE(0xB0, 0x6)	},
	{ SMCA_L3_CACHE, HWID_MCATYPE(0xB0, 0x7)	},

	/* Data Fabric MCA types */
	{ SMCA_CS,	 HWID_MCATYPE(0x2E, 0x0)	},
	{ SMCA_PIE,	 HWID_MCATYPE(0x2E, 0x1)	},
	{ SMCA_CS_V2,	 HWID_MCATYPE(0x2E, 0x2)	},
	{ SMCA_MA_LLC,	 HWID_MCATYPE(0x2E, 0x4)	},

	/* Unified Memory Controller MCA type */
	{ SMCA_UMC,	 HWID_MCATYPE(0x96, 0x0)	},
	{ SMCA_UMC_V2,	 HWID_MCATYPE(0x96, 0x1)	},

	/* Parameter Block MCA type */
	{ SMCA_PB,	 HWID_MCATYPE(0x05, 0x0)	},

	/* Platform Security Processor MCA type */
	{ SMCA_PSP,	 HWID_MCATYPE(0xFF, 0x0)	},
	{ SMCA_PSP_V2,	 HWID_MCATYPE(0xFF, 0x1)	},

	/* System Management Unit MCA type */
	{ SMCA_SMU,	 HWID_MCATYPE(0x01, 0x0)	},
	{ SMCA_SMU_V2,	 HWID_MCATYPE(0x01, 0x1)	},

	/* Microprocessor 5 Unit MCA type */
	{ SMCA_MP5,	 HWID_MCATYPE(0x01, 0x2)	},

	/* MPDMA MCA type */
	{ SMCA_MPDMA,	 HWID_MCATYPE(0x01, 0x3)	},

	/* Northbridge IO Unit MCA type */
	{ SMCA_NBIO,	 HWID_MCATYPE(0x18, 0x0)	},

	/* PCI Express Unit MCA type */
	{ SMCA_PCIE,	 HWID_MCATYPE(0x46, 0x0)	},
	{ SMCA_PCIE_V2,	 HWID_MCATYPE(0x46, 0x1)	},

	{ SMCA_XGMI_PCS, HWID_MCATYPE(0x50, 0x0)	},
	{ SMCA_NBIF,	 HWID_MCATYPE(0x6C, 0x0)	},
	{ SMCA_SHUB,	 HWID_MCATYPE(0x80, 0x0)	},
	{ SMCA_SATA,	 HWID_MCATYPE(0xA8, 0x0)	},
	{ SMCA_USB,	 HWID_MCATYPE(0xAA, 0x0)	},
	{ SMCA_USR_DP,	 HWID_MCATYPE(0x170, 0x0)	},
	{ SMCA_USR_CP,	 HWID_MCATYPE(0x180, 0x0)	},
	{ SMCA_GMI_PCS,  HWID_MCATYPE(0x241, 0x0)	},
	{ SMCA_XGMI_PHY, HWID_MCATYPE(0x259, 0x0)	},
	{ SMCA_WAFL_PHY, HWID_MCATYPE(0x267, 0x0)	},
	{ SMCA_GMI_PHY,	 HWID_MCATYPE(0x269, 0x0)	},
};

/*
 * In SMCA enabled processors, we can have multiple banks for a given IP type.
 * So to define a unique name for each bank, we use a temp c-string to append
 * the MCA_IPID[InstanceId] to type's name in get_name().
 *
 * InstanceId is 32 bits which is 8 characters. Make sure MAX_MCATYPE_NAME_LEN
 * is greater than 8 plus 1 (for underscore) plus length of longest type name.
 */
#define MAX_MCATYPE_NAME_LEN	30
static char buf_mcatype[MAX_MCATYPE_NAME_LEN];

struct threshold_block {
	/* This block's number within its bank. */
	unsigned int		block;
	/* MCA bank number that contains this block. */
	unsigned int		bank;
	/* CPU which controls this block's MCA bank. */
	unsigned int		cpu;
	/* MCA_MISC MSR address for this block. */
	u32			address;
	/* Enable/Disable APIC interrupt. */
	bool			interrupt_enable;
	/* Bank can generate an interrupt. */
	bool			interrupt_capable;
	/* Value upon which threshold interrupt is generated. */
	u16			threshold_limit;
	/* sysfs object */
	struct kobject		kobj;
	/* List of threshold blocks within this block's MCA bank. */
	struct list_head	miscj;
};

struct threshold_bank {
	struct kobject		*kobj;
	/* List of threshold blocks within this MCA bank. */
	struct list_head	miscj;
};

static DEFINE_PER_CPU(struct threshold_bank **, threshold_banks);

/*
 * A list of the banks enabled on each logical CPU. Controls which respective
 * descriptors to initialize later in mce_threshold_create_device().
 */
static DEFINE_PER_CPU(u64, bank_map);

static void amd_threshold_interrupt(void);
static void amd_deferred_error_interrupt(void);

static void default_deferred_error_interrupt(void)
{
	pr_err("Unexpected deferred interrupt at vector %x\n", DEFERRED_ERROR_VECTOR);
}
void (*deferred_error_int_vector)(void) = default_deferred_error_interrupt;

static void smca_configure(unsigned int bank, unsigned int cpu)
{
	struct mce_amd_cpu_data *data = this_cpu_ptr(&mce_amd_data);
	u8 *bank_counts = this_cpu_ptr(smca_bank_counts);
	const struct smca_hwid *s_hwid;
	unsigned int i, hwid_mcatype;
	u32 high, low;
	u32 smca_config = MSR_AMD64_SMCA_MCx_CONFIG(bank);

	/* Set appropriate bits in MCA_CONFIG */
	if (!rdmsr_safe(smca_config, &low, &high)) {
		/*
		 * OS is required to set the MCAX bit to acknowledge that it is
		 * now using the new MSR ranges and new registers under each
		 * bank. It also means that the OS will configure deferred
		 * errors in the new MCx_CONFIG register. If the bit is not set,
		 * uncorrectable errors will cause a system panic.
		 *
		 * MCA_CONFIG[MCAX] is bit 32 (0 in the high portion of the MSR.)
		 */
		high |= BIT(0);

		/*
		 * SMCA sets the Deferred Error Interrupt type per bank.
		 *
		 * MCA_CONFIG[DeferredIntTypeSupported] is bit 5, and tells us
		 * if the DeferredIntType bit field is available.
		 *
		 * MCA_CONFIG[DeferredIntType] is bits [38:37] ([6:5] in the
		 * high portion of the MSR). OS should set this to 0x1 to enable
		 * APIC based interrupt. First, check that no interrupt has been
		 * set.
		 */
		if ((low & BIT(5)) && !((high >> 5) & 0x3) && data->dfr_intr_en) {
			__set_bit(bank, data->dfr_intr_banks);
			high |= BIT(5);
		}

		/*
		 * SMCA Corrected Error Interrupt
		 *
		 * MCA_CONFIG[IntPresent] is bit 10, and tells us if the bank can
		 * send an MCA Thresholding interrupt without the OS initializing
		 * this feature. This can be used if the threshold limit is managed
		 * by the platform.
		 *
		 * MCA_CONFIG[IntEn] is bit 40 (8 in the high portion of the MSR).
		 * The OS should set this to inform the platform that the OS is ready
		 * to handle the MCA Thresholding interrupt.
		 */
		if ((low & BIT(10)) && data->thr_intr_en) {
			__set_bit(bank, data->thr_intr_banks);
			high |= BIT(8);
		}

		this_cpu_ptr(mce_banks_array)[bank].lsb_in_status = !!(low & BIT(8));

		if (low & MCI_CONFIG_PADDRV)
			this_cpu_ptr(smca_banks)[bank].paddrv = 1;

		wrmsr(smca_config, low, high);
	}

	if (rdmsr_safe(MSR_AMD64_SMCA_MCx_IPID(bank), &low, &high)) {
		pr_warn("Failed to read MCA_IPID for bank %d\n", bank);
		return;
	}

	hwid_mcatype = HWID_MCATYPE(high & MCI_IPID_HWID,
				    (high & MCI_IPID_MCATYPE) >> 16);

	for (i = 0; i < ARRAY_SIZE(smca_hwid_mcatypes); i++) {
		s_hwid = &smca_hwid_mcatypes[i];

		if (hwid_mcatype == s_hwid->hwid_mcatype) {
			this_cpu_ptr(smca_banks)[bank].hwid = s_hwid;
			this_cpu_ptr(smca_banks)[bank].id = low;
			this_cpu_ptr(smca_banks)[bank].sysfs_id = bank_counts[s_hwid->bank_type]++;
			break;
		}
	}
}

struct thresh_restart {
	struct threshold_block	*b;
	int			set_lvt_off;
	int			lvt_off;
	u16			old_limit;
};

static const char *bank4_names(const struct threshold_block *b)
{
	switch (b->address) {
	/* MSR4_MISC0 */
	case 0x00000413:
		return "dram";

	case 0xc0000408:
		return "ht_links";

	case 0xc0000409:
		return "l3_cache";

	default:
		WARN(1, "Funny MSR: 0x%08x\n", b->address);
		return "";
	}
};


static bool lvt_interrupt_supported(unsigned int bank, u32 msr_high_bits)
{
	/*
	 * bank 4 supports APIC LVT interrupts implicitly since forever.
	 */
	if (bank == 4)
		return true;

	/*
	 * IntP: interrupt present; if this bit is set, the thresholding
	 * bank can generate APIC LVT interrupts
	 */
	return msr_high_bits & BIT(28);
}

static bool lvt_off_valid(struct threshold_block *b, int apic, u32 lo, u32 hi)
{
	int msr = (hi & MASK_LVTOFF_HI) >> 20;

	/*
	 * On SMCA CPUs, LVT offset is programmed at a different MSR, and
	 * the BIOS provides the value. The original field where LVT offset
	 * was set is reserved. Return early here:
	 */
	if (mce_flags.smca)
		return false;

	if (apic < 0) {
		pr_err(FW_BUG "cpu %d, failed to setup threshold interrupt "
		       "for bank %d, block %d (MSR%08X=0x%x%08x)\n", b->cpu,
		       b->bank, b->block, b->address, hi, lo);
		return false;
	}

	if (apic != msr) {
		pr_err(FW_BUG "cpu %d, invalid threshold interrupt offset %d "
		       "for bank %d, block %d (MSR%08X=0x%x%08x)\n",
		       b->cpu, apic, b->bank, b->block, b->address, hi, lo);
		return false;
	}

	return true;
};

/* Reprogram MCx_MISC MSR behind this threshold block. */
static void threshold_restart_block(void *_tr)
{
	struct thresh_restart *tr = _tr;
	u32 hi, lo;

	/* sysfs write might race against an offline operation */
	if (!this_cpu_read(threshold_banks) && !tr->set_lvt_off)
		return;

	rdmsr(tr->b->address, lo, hi);

	/*
	 * Reset error count and overflow bit.
	 * This is done during init or after handling an interrupt.
	 */
	if (hi & MASK_OVERFLOW_HI || tr->set_lvt_off) {
		hi &= ~(MASK_ERR_COUNT_HI | MASK_OVERFLOW_HI);
		hi |= THRESHOLD_MAX - tr->b->threshold_limit;
	} else if (tr->old_limit) {	/* change limit w/o reset */
		int new_count = (hi & THRESHOLD_MAX) +
		    (tr->old_limit - tr->b->threshold_limit);

		hi = (hi & ~MASK_ERR_COUNT_HI) |
		    (new_count & THRESHOLD_MAX);
	}

	/* clear IntType */
	hi &= ~MASK_INT_TYPE_HI;

	if (!tr->b->interrupt_capable)
		goto done;

	if (tr->set_lvt_off) {
		if (lvt_off_valid(tr->b, tr->lvt_off, lo, hi)) {
			/* set new lvt offset */
			hi &= ~MASK_LVTOFF_HI;
			hi |= tr->lvt_off << 20;
		}
	}

	if (tr->b->interrupt_enable)
		hi |= INT_TYPE_APIC;

 done:

	hi |= MASK_COUNT_EN_HI;
	wrmsr(tr->b->address, lo, hi);
}

static void threshold_restart_bank(unsigned int bank, bool intr_en)
{
	struct threshold_bank **thr_banks = this_cpu_read(threshold_banks);
	struct threshold_block *block, *tmp;
	struct thresh_restart tr;

	if (!thr_banks || !thr_banks[bank])
		return;

	memset(&tr, 0, sizeof(tr));

	list_for_each_entry_safe(block, tmp, &thr_banks[bank]->miscj, miscj) {
		tr.b = block;
		tr.b->interrupt_enable = intr_en;
		threshold_restart_block(&tr);
	}
}

/* Try to use the threshold limit reported through APEI. */
static u16 get_thr_limit(void)
{
	u32 thr_limit = mce_get_apei_thr_limit();

	/* Fallback to old default if APEI limit is not available. */
	if (!thr_limit)
		return THRESHOLD_MAX;

	return min(thr_limit, THRESHOLD_MAX);
}

static void mce_threshold_block_init(struct threshold_block *b, int offset)
{
	struct thresh_restart tr = {
		.b			= b,
		.set_lvt_off		= 1,
		.lvt_off		= offset,
	};

	b->threshold_limit		= get_thr_limit();
	threshold_restart_block(&tr);
};

static int setup_APIC_mce_threshold(int reserved, int new)
{
	if (reserved < 0 && !setup_APIC_eilvt(new, THRESHOLD_APIC_VECTOR,
					      APIC_EILVT_MSG_FIX, 0))
		return new;

	return reserved;
}

static u32 get_block_address(u32 current_addr, u32 low, u32 high,
			     unsigned int bank, unsigned int block,
			     unsigned int cpu)
{
	u32 addr = 0, offset = 0;

	if ((bank >= per_cpu(mce_num_banks, cpu)) || (block >= NR_BLOCKS))
		return addr;

	if (mce_flags.smca) {
		if (!block)
			return MSR_AMD64_SMCA_MCx_MISC(bank);

		if (!(low & MASK_BLKPTR_LO))
			return 0;

		return MSR_AMD64_SMCA_MCx_MISCy(bank, block - 1);
	}

	/* Fall back to method we used for older processors: */
	switch (block) {
	case 0:
		addr = mca_msr_reg(bank, MCA_MISC);
		break;
	case 1:
		offset = ((low & MASK_BLKPTR_LO) >> 21);
		if (offset)
			addr = MCG_XBLK_ADDR + offset;
		break;
	default:
		addr = ++current_addr;
	}
	return addr;
}

static int prepare_threshold_block(unsigned int bank, unsigned int block, u32 addr,
				   int offset, u32 misc_high)
{
	unsigned int cpu = smp_processor_id();
	struct threshold_block b;
	int new;

	if (!block)
		per_cpu(bank_map, cpu) |= BIT_ULL(bank);

	memset(&b, 0, sizeof(b));
	b.cpu			= cpu;
	b.bank			= bank;
	b.block			= block;
	b.address		= addr;
	b.interrupt_capable	= lvt_interrupt_supported(bank, misc_high);

	if (!b.interrupt_capable)
		goto done;

	__set_bit(bank, this_cpu_ptr(&mce_amd_data)->thr_intr_banks);
	b.interrupt_enable = 1;

	if (mce_flags.smca)
		goto done;

	new = (misc_high & MASK_LVTOFF_HI) >> 20;
	offset = setup_APIC_mce_threshold(offset, new);
	if (offset == new)
		thresholding_irq_en = true;

done:
	mce_threshold_block_init(&b, offset);

	return offset;
}

bool amd_filter_mce(struct mce *m)
{
	enum smca_bank_types bank_type = smca_get_bank_type(m->extcpu, m->bank);
	struct cpuinfo_x86 *c = &boot_cpu_data;

	/* See Family 17h Models 10h-2Fh Erratum #1114. */
	if (c->x86 == 0x17 &&
	    c->x86_model >= 0x10 && c->x86_model <= 0x2F &&
	    bank_type == SMCA_IF && XEC(m->status, 0x3f) == 10)
		return true;

	/* NB GART TLB error reporting is disabled by default. */
	if (c->x86 < 0x17) {
		if (m->bank == 4 && XEC(m->status, 0x1f) == 0x5)
			return true;
	}

	return false;
}

/*
 * Turn off thresholding banks for the following conditions:
 * - MC4_MISC thresholding is not supported on Family 0x15.
 * - Prevent possible spurious interrupts from the IF bank on Family 0x17
 *   Models 0x10-0x2F due to Erratum #1114.
 */
static void disable_err_thresholding(struct cpuinfo_x86 *c, unsigned int bank)
{
	int i, num_msrs;
	u64 hwcr;
	bool need_toggle;
	u32 msrs[NR_BLOCKS];

	if (c->x86 == 0x15 && bank == 4) {
		msrs[0] = 0x00000413; /* MC4_MISC0 */
		msrs[1] = 0xc0000408; /* MC4_MISC1 */
		num_msrs = 2;
	} else if (c->x86 == 0x17 &&
		   (c->x86_model >= 0x10 && c->x86_model <= 0x2F)) {

		if (smca_get_bank_type(smp_processor_id(), bank) != SMCA_IF)
			return;

		msrs[0] = MSR_AMD64_SMCA_MCx_MISC(bank);
		num_msrs = 1;
	} else {
		return;
	}

	rdmsrq(MSR_K7_HWCR, hwcr);

	/* McStatusWrEn has to be set */
	need_toggle = !(hwcr & BIT(18));
	if (need_toggle)
		wrmsrq(MSR_K7_HWCR, hwcr | BIT(18));

	/* Clear CntP bit safely */
	for (i = 0; i < num_msrs; i++)
		msr_clear_bit(msrs[i], 62);

	/* restore old settings */
	if (need_toggle)
		wrmsrq(MSR_K7_HWCR, hwcr);
}

static void amd_apply_cpu_quirks(struct cpuinfo_x86 *c)
{
	struct mce_bank *mce_banks = this_cpu_ptr(mce_banks_array);

	/* This should be disabled by the BIOS, but isn't always */
	if (c->x86 == 15 && this_cpu_read(mce_num_banks) > 4) {
		/*
		 * disable GART TBL walk error reporting, which
		 * trips off incorrectly with the IOMMU & 3ware
		 * & Cerberus:
		 */
		clear_bit(10, (unsigned long *)&mce_banks[4].ctl);
	}

	/*
	 * Various K7s with broken bank 0 around. Always disable
	 * by default.
	 */
	if (c->x86 == 6 && this_cpu_read(mce_num_banks))
		mce_banks[0].ctl = 0;
}

/*
 * Enable the APIC LVT interrupt vectors once per-CPU. This should be done before hardware is
 * ready to send interrupts.
 *
 * Individual error sources are enabled later during per-bank init.
 */
static void smca_enable_interrupt_vectors(void)
{
	struct mce_amd_cpu_data *data = this_cpu_ptr(&mce_amd_data);
	u64 mca_intr_cfg, offset;

	if (!mce_flags.smca || !mce_flags.succor)
		return;

	if (rdmsrq_safe(MSR_CU_DEF_ERR, &mca_intr_cfg))
		return;

	offset = (mca_intr_cfg & SMCA_THR_LVT_OFF) >> 12;
	if (!setup_APIC_eilvt(offset, THRESHOLD_APIC_VECTOR, APIC_EILVT_MSG_FIX, 0))
		data->thr_intr_en = 1;

	offset = (mca_intr_cfg & MASK_DEF_LVTOFF) >> 4;
	if (!setup_APIC_eilvt(offset, DEFERRED_ERROR_VECTOR, APIC_EILVT_MSG_FIX, 0))
		data->dfr_intr_en = 1;
}

/* cpu init entry point, called from mce.c with preempt off */
void mce_amd_feature_init(struct cpuinfo_x86 *c)
{
	unsigned int bank, block, cpu = smp_processor_id();
	u32 low = 0, high = 0, address = 0;
	int offset = -1;

	amd_apply_cpu_quirks(c);

	mce_flags.amd_threshold	 = 1;

	smca_enable_interrupt_vectors();

	for (bank = 0; bank < this_cpu_read(mce_num_banks); ++bank) {
		if (mce_flags.smca) {
			smca_configure(bank, cpu);

			if (!this_cpu_ptr(&mce_amd_data)->thr_intr_en)
				continue;
		}

		disable_err_thresholding(c, bank);

		for (block = 0; block < NR_BLOCKS; ++block) {
			address = get_block_address(address, low, high, bank, block, cpu);
			if (!address)
				break;

			if (rdmsr_safe(address, &low, &high))
				break;

			if (!(high & MASK_VALID_HI))
				continue;

			if (!(high & MASK_CNTP_HI)  ||
			     (high & MASK_LOCKED_HI))
				continue;

			offset = prepare_threshold_block(bank, block, address, offset, high);
		}
	}
}

void smca_bsp_init(void)
{
	mce_threshold_vector	  = amd_threshold_interrupt;
	deferred_error_int_vector = amd_deferred_error_interrupt;
}

/*
 * DRAM ECC errors are reported in the Northbridge (bank 4) with
 * Extended Error Code 8.
 */
static bool legacy_mce_is_memory_error(struct mce *m)
{
	return m->bank == 4 && XEC(m->status, 0x1f) == 8;
}

/*
 * DRAM ECC errors are reported in Unified Memory Controllers with
 * Extended Error Code 0.
 */
static bool smca_mce_is_memory_error(struct mce *m)
{
	enum smca_bank_types bank_type;

	if (XEC(m->status, 0x3f))
		return false;

	bank_type = smca_get_bank_type(m->extcpu, m->bank);

	return bank_type == SMCA_UMC || bank_type == SMCA_UMC_V2;
}

bool amd_mce_is_memory_error(struct mce *m)
{
	if (mce_flags.smca)
		return smca_mce_is_memory_error(m);
	else
		return legacy_mce_is_memory_error(m);
}

/*
 * Some AMD systems have an explicit indicator that the value in MCA_ADDR is a
 * system physical address. Individual cases though, need to be detected for
 * other systems. Future cases will be added as needed.
 *
 * 1) General case
 *	a) Assume address is not usable.
 * 2) Poison errors
 *	a) Indicated by MCA_STATUS[43]: poison. Defined for all banks except legacy
 *	   northbridge (bank 4).
 *	b) Refers to poison consumption in the core. Does not include "no action",
 *	   "action optional", or "deferred" error severities.
 *	c) Will include a usable address so that immediate action can be taken.
 * 3) Northbridge DRAM ECC errors
 *	a) Reported in legacy bank 4 with extended error code (XEC) 8.
 *	b) MCA_STATUS[43] is *not* defined as poison in legacy bank 4. Therefore,
 *	   this bit should not be checked.
 * 4) MCI_STATUS_PADDRVAL is set
 *	a) Will provide a valid system physical address.
 *
 * NOTE: SMCA UMC memory errors fall into case #1.
 */
bool amd_mce_usable_address(struct mce *m)
{
	/* Check special northbridge case 3) first. */
	if (!mce_flags.smca) {
		if (legacy_mce_is_memory_error(m))
			return true;
		else if (m->bank == 4)
			return false;
	}

	if (this_cpu_ptr(smca_banks)[m->bank].paddrv)
		return m->status & MCI_STATUS_PADDRV;

	/* Check poison bit for all other bank types. */
	if (m->status & MCI_STATUS_POISON)
		return true;

	/* Assume address is not usable for all others. */
	return false;
}

DEFINE_IDTENTRY_SYSVEC(sysvec_deferred_error)
{
	trace_deferred_error_apic_entry(DEFERRED_ERROR_VECTOR);
	inc_irq_stat(irq_deferred_error_count);
	deferred_error_int_vector();
	trace_deferred_error_apic_exit(DEFERRED_ERROR_VECTOR);
	apic_eoi();
}

/* APIC interrupt handler for deferred errors */
static void amd_deferred_error_interrupt(void)
{
	machine_check_poll(MCP_TIMESTAMP, &this_cpu_ptr(&mce_amd_data)->dfr_intr_banks);
}

void mce_amd_handle_storm(unsigned int bank, bool on)
{
	threshold_restart_bank(bank, on);
}

static void amd_reset_thr_limit(unsigned int bank)
{
	threshold_restart_bank(bank, true);
}

/*
 * Threshold interrupt handler will service THRESHOLD_APIC_VECTOR. The interrupt
 * goes off when error_count reaches threshold_limit.
 */
static void amd_threshold_interrupt(void)
{
	machine_check_poll(MCP_TIMESTAMP, &this_cpu_ptr(&mce_amd_data)->thr_intr_banks);
}

void amd_clear_bank(struct mce *m)
{
	amd_reset_thr_limit(m->bank);

	/* Clear MCA_DESTAT for all deferred errors even those logged in MCA_STATUS. */
	if (m->status & MCI_STATUS_DEFERRED)
		mce_wrmsrq(MSR_AMD64_SMCA_MCx_DESTAT(m->bank), 0);

	/* Don't clear MCA_STATUS if MCA_DESTAT was used exclusively. */
	if (m->kflags & MCE_CHECK_DFR_REGS)
		return;

	mce_wrmsrq(mca_msr_reg(m->bank, MCA_STATUS), 0);
}

/*
 * Sysfs Interface
 */

struct threshold_attr {
	struct attribute attr;
	ssize_t (*show) (struct threshold_block *, char *);
	ssize_t (*store) (struct threshold_block *, const char *, size_t count);
};

#define SHOW_FIELDS(name)						\
static ssize_t show_ ## name(struct threshold_block *b, char *buf)	\
{									\
	return sprintf(buf, "%lu\n", (unsigned long) b->name);		\
}
SHOW_FIELDS(interrupt_enable)
SHOW_FIELDS(threshold_limit)

static ssize_t
store_interrupt_enable(struct threshold_block *b, const char *buf, size_t size)
{
	struct thresh_restart tr;
	unsigned long new;

	if (!b->interrupt_capable)
		return -EINVAL;

	if (kstrtoul(buf, 0, &new) < 0)
		return -EINVAL;

	b->interrupt_enable = !!new;

	memset(&tr, 0, sizeof(tr));
	tr.b		= b;

	if (smp_call_function_single(b->cpu, threshold_restart_block, &tr, 1))
		return -ENODEV;

	return size;
}

static ssize_t
store_threshold_limit(struct threshold_block *b, const char *buf, size_t size)
{
	struct thresh_restart tr;
	unsigned long new;

	if (kstrtoul(buf, 0, &new) < 0)
		return -EINVAL;

	if (new > THRESHOLD_MAX)
		new = THRESHOLD_MAX;
	if (new < 1)
		new = 1;

	memset(&tr, 0, sizeof(tr));
	tr.old_limit = b->threshold_limit;
	b->threshold_limit = new;
	tr.b = b;

	if (smp_call_function_single(b->cpu, threshold_restart_block, &tr, 1))
		return -ENODEV;

	return size;
}

static ssize_t show_error_count(struct threshold_block *b, char *buf)
{
	u32 lo, hi;

	/* CPU might be offline by now */
	if (rdmsr_on_cpu(b->cpu, b->address, &lo, &hi))
		return -ENODEV;

	return sprintf(buf, "%u\n", ((hi & THRESHOLD_MAX) -
				     (THRESHOLD_MAX - b->threshold_limit)));
}

static struct threshold_attr error_count = {
	.attr = {.name = __stringify(error_count), .mode = 0444 },
	.show = show_error_count,
};

#define RW_ATTR(val)							\
static struct threshold_attr val = {					\
	.attr	= {.name = __stringify(val), .mode = 0644 },		\
	.show	= show_## val,						\
	.store	= store_## val,						\
};

RW_ATTR(interrupt_enable);
RW_ATTR(threshold_limit);

static struct attribute *default_attrs[] = {
	&threshold_limit.attr,
	&error_count.attr,
	NULL,	/* possibly interrupt_enable if supported, see below */
	NULL,
};
ATTRIBUTE_GROUPS(default);

#define to_block(k)	container_of(k, struct threshold_block, kobj)
#define to_attr(a)	container_of(a, struct threshold_attr, attr)

static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
{
	struct threshold_block *b = to_block(kobj);
	struct threshold_attr *a = to_attr(attr);
	ssize_t ret;

	ret = a->show ? a->show(b, buf) : -EIO;

	return ret;
}

static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
{
	struct threshold_block *b = to_block(kobj);
	struct threshold_attr *a = to_attr(attr);
	ssize_t ret;

	ret = a->store ? a->store(b, buf, count) : -EIO;

	return ret;
}

static const struct sysfs_ops threshold_ops = {
	.show			= show,
	.store			= store,
};

static void threshold_block_release(struct kobject *kobj);

static const struct kobj_type threshold_ktype = {
	.sysfs_ops		= &threshold_ops,
	.default_groups		= default_groups,
	.release		= threshold_block_release,
};

static const char *get_name(unsigned int cpu, unsigned int bank, struct threshold_block *b)
{
	enum smca_bank_types bank_type;

	if (!mce_flags.smca) {
		if (b && bank == 4)
			return bank4_names(b);

		return th_names[bank];
	}

	bank_type = smca_get_bank_type(cpu, bank);

	if (b && (bank_type == SMCA_UMC || bank_type == SMCA_UMC_V2)) {
		if (b->block < ARRAY_SIZE(smca_umc_block_names))
			return smca_umc_block_names[b->block];
	}

	if (b && b->block) {
		snprintf(buf_mcatype, MAX_MCATYPE_NAME_LEN, "th_block_%u", b->block);
		return buf_mcatype;
	}

	if (bank_type >= N_SMCA_BANK_TYPES) {
		snprintf(buf_mcatype, MAX_MCATYPE_NAME_LEN, "th_bank_%u", bank);
		return buf_mcatype;
	}

	if (per_cpu(smca_bank_counts, cpu)[bank_type] == 1)
		return smca_get_name(bank_type);

	snprintf(buf_mcatype, MAX_MCATYPE_NAME_LEN,
		 "%s_%u", smca_get_name(bank_type),
			  per_cpu(smca_banks, cpu)[bank].sysfs_id);
	return buf_mcatype;
}

static int allocate_threshold_blocks(unsigned int cpu, struct threshold_bank *tb,
				     unsigned int bank, unsigned int block,
				     u32 address)
{
	struct threshold_block *b = NULL;
	u32 low, high;
	int err;

	if ((bank >= this_cpu_read(mce_num_banks)) || (block >= NR_BLOCKS))
		return 0;

	if (rdmsr_safe(address, &low, &high))
		return 0;

	if (!(high & MASK_VALID_HI)) {
		if (block)
			goto recurse;
		else
			return 0;
	}

	if (!(high & MASK_CNTP_HI)  ||
	     (high & MASK_LOCKED_HI))
		goto recurse;

	b = kzalloc(sizeof(struct threshold_block), GFP_KERNEL);
	if (!b)
		return -ENOMEM;

	b->block		= block;
	b->bank			= bank;
	b->cpu			= cpu;
	b->address		= address;
	b->interrupt_enable	= 0;
	b->interrupt_capable	= lvt_interrupt_supported(bank, high);
	b->threshold_limit	= get_thr_limit();

	if (b->interrupt_capable) {
		default_attrs[2] = &interrupt_enable.attr;
		b->interrupt_enable = 1;
	} else {
		default_attrs[2] = NULL;
	}

	list_add(&b->miscj, &tb->miscj);

	mce_threshold_block_init(b, (high & MASK_LVTOFF_HI) >> 20);

	err = kobject_init_and_add(&b->kobj, &threshold_ktype, tb->kobj, get_name(cpu, bank, b));
	if (err)
		goto out_free;
recurse:
	address = get_block_address(address, low, high, bank, ++block, cpu);
	if (!address)
		return 0;

	err = allocate_threshold_blocks(cpu, tb, bank, block, address);
	if (err)
		goto out_free;

	if (b)
		kobject_uevent(&b->kobj, KOBJ_ADD);

	return 0;

out_free:
	if (b) {
		list_del(&b->miscj);
		kobject_put(&b->kobj);
	}
	return err;
}

static int threshold_create_bank(struct threshold_bank **bp, unsigned int cpu,
				 unsigned int bank)
{
	struct device *dev = this_cpu_read(mce_device);
	struct threshold_bank *b = NULL;
	const char *name = get_name(cpu, bank, NULL);
	int err = 0;

	if (!dev)
		return -ENODEV;

	b = kzalloc(sizeof(struct threshold_bank), GFP_KERNEL);
	if (!b) {
		err = -ENOMEM;
		goto out;
	}

	/* Associate the bank with the per-CPU MCE device */
	b->kobj = kobject_create_and_add(name, &dev->kobj);
	if (!b->kobj) {
		err = -EINVAL;
		goto out_free;
	}

	INIT_LIST_HEAD(&b->miscj);

	err = allocate_threshold_blocks(cpu, b, bank, 0, mca_msr_reg(bank, MCA_MISC));
	if (err)
		goto out_kobj;

	bp[bank] = b;
	return 0;

out_kobj:
	kobject_put(b->kobj);
out_free:
	kfree(b);
out:
	return err;
}

static void threshold_block_release(struct kobject *kobj)
{
	kfree(to_block(kobj));
}

static void threshold_remove_bank(struct threshold_bank *bank)
{
	struct threshold_block *pos, *tmp;

	list_for_each_entry_safe(pos, tmp, &bank->miscj, miscj) {
		list_del(&pos->miscj);
		kobject_put(&pos->kobj);
	}

	kobject_put(bank->kobj);
	kfree(bank);
}

static void __threshold_remove_device(struct threshold_bank **bp)
{
	unsigned int bank, numbanks = this_cpu_read(mce_num_banks);

	for (bank = 0; bank < numbanks; bank++) {
		if (!bp[bank])
			continue;

		threshold_remove_bank(bp[bank]);
		bp[bank] = NULL;
	}
	kfree(bp);
}

void mce_threshold_remove_device(unsigned int cpu)
{
	struct threshold_bank **bp = this_cpu_read(threshold_banks);

	if (!bp)
		return;

	/*
	 * Clear the pointer before cleaning up, so that the interrupt won't
	 * touch anything of this.
	 */
	this_cpu_write(threshold_banks, NULL);

	__threshold_remove_device(bp);
	return;
}

/**
 * mce_threshold_create_device - Create the per-CPU MCE threshold device
 * @cpu:	The plugged in CPU
 *
 * Create directories and files for all valid threshold banks.
 *
 * This is invoked from the CPU hotplug callback which was installed in
 * mcheck_init_device(). The invocation happens in context of the hotplug
 * thread running on @cpu.  The callback is invoked on all CPUs which are
 * online when the callback is installed or during a real hotplug event.
 */
void mce_threshold_create_device(unsigned int cpu)
{
	unsigned int numbanks, bank;
	struct threshold_bank **bp;

	if (!mce_flags.amd_threshold)
		return;

	bp = this_cpu_read(threshold_banks);
	if (bp)
		return;

	numbanks = this_cpu_read(mce_num_banks);
	bp = kcalloc(numbanks, sizeof(*bp), GFP_KERNEL);
	if (!bp)
		return;

	for (bank = 0; bank < numbanks; ++bank) {
		if (!(this_cpu_read(bank_map) & BIT_ULL(bank)))
			continue;
		if (threshold_create_bank(bp, cpu, bank)) {
			__threshold_remove_device(bp);
			return;
		}
	}
	this_cpu_write(threshold_banks, bp);

	if (thresholding_irq_en)
		mce_threshold_vector = amd_threshold_interrupt;
	return;
}